# GenderVL-Bench

**GenderVL-Bench** is a compact vision-language benchmark for evaluating how Vision-Language Models (VLMs) interpret **gender-related representations across different occupations**.

### Dataset

* **108 images**
* **12 occupations**
* **9 images per occupation**
* **Format:** JPEG / ImageFolder
* **Split:** `train`

### Usage

```python
from datasets import load_dataset

dataset = load_dataset("suparnojit/GenderVL-Bench")
```

### Citation

```bibtex
@dataset{sarkar2026gendervlbench,
  author    = {Sarkar, Suparnojit},
  title     = {GenderVL-Bench},
  year      = {2026},
  publisher = {Hugging Face},
  url       = {https://huggingface.co/datasets/suparnojit/GenderVL-Bench}
}
```